Why a National Patient Identifier won’t solve matching concerns
Hope for a national patient identifier suffered a blow recently when the College of Healthcare Information Management Executives announced it was pulling the plug on its highly publicized National Patient ID Challenge.
For years, the question of how to accurately identify individuals in a complicated healthcare ecosystem has been a perplexing one for healthcare IT leaders.
Data creation in healthcare has accelerated from torrential to a veritable tsunami, bringing with it many challenges in the identification and reconciliation of patient records because of the various ways systems classify, store and protect information. As data sharing proliferates and the bar for better coordination across the continuum is raised, resolving patient record matching issues grows especially complex—and increasingly urgent.
Fragmented data trapped in silos makes tracking patients across care settings especially difficult, leading to increased risk and cost from redundant medical tests and procedures. On average, 8 to 12 percent of a health organization’s patient records are duplicates and the expense of repeated medical care because of duplicate records can cost $1,100 per patient, studies estimate.
Duplicate and disjointed records continue to put patient safety at risk and plague providers’ bottom lines. Without a unique way of identifying each patient, providers will continue to struggle to meet the growing demands of value-based care. With no foreseeable action by Congress, the industry must ask whether establishing an NPI is worth the investment. And is it really the silver bullet in solving the patient record-matching issues?
The need for an NPI has been part of policy discussions since Congress introduced the Health Insurance Portability and Accountability Act (HIPAA) in 1996. In fact, HIPAA’s architects realized that a unique patient identifier was necessary for accurate data sharing, but privacy concerns quickly quashed any NPI progress.
Detractors feared that tying all of an individual’s health information to a single identifier could prove disastrous. A slate of recent high-profile breaches involving effecting massive numbers of people has only reinforced this concern. Despite increasingly sophisticated security protocols, privacy is still the primary barrier to political support.
When Congress dismissed the concept of the NPI in 1998, the industry really didn’t have the technology in place to facilitate widespread exchange of patient data anyway. Today, however, in the wake of digitization, healthcare organizations are inundated with patient data. Although the infrastructure is available, transmission and capture of a single patient identifier across thousands of disparate systems could easily take decades for full adoption.
In terms of cost for the required technology, efforts to institute a unique patient identifier in the United Kingdom have demonstrated that a national infrastructure for managing information costs billions of dollars.
The UK government’s efforts to implement a national patient identifier serves as an instructive case study about the complexities involved in establishing a nationwide program—even in a country with a nationalized healthcare system.
In 1996, the National Health Service (NHS) introduced a unique patient identifier, called the NHS number. After 20 years of intense IT labor and 10 billion pounds in expenses, the NHS has turned a corner in implementing a single patient identifier, although it is by no means a perfect mechanism of patient identification.
While the mandated NHS number is a positive move, it alone is still not enough to achieve complete integration of information across health and social care services. For example, some older health system applications still cannot connect real time to query for the NHS number. Moreover, health systems continue to encounter inaccuracies in demographics; multiple NHS numbers being applied to a single patient; and incorrect numbers applied to patients. In fact, even with NHS numbers in place, leaders have discovered that an external enterprise master patient index (EMPI) solution is needed to help manage identifiers.
Most recently, the NHS has turned its attention to introducing the unique number into social care systems that provide services for patients after discharge. Leaders estimate it will take 10 to 15 years to get these systems (which were never designed to manage an NHS number) integrated into the program. In the interim, an external EMPI will be necessary as a stopgap measure.
Further, Scotland is replacing its equivalent, the Community Health Index, with an EMPI to assign and manage unique patient identifiers across agencies and service providers. As part of the Mainframe Solutions Transformation Programme (MSTP), the EMPI will become the issuing authority for the CHI number, a number critical to the electronic patient record across Scotland.
The U.S. healthcare industry is much larger in scope and far more complex than that of the UK, and thus the difficulties in implementing an NPI can be expected to be proportionally greater. That is not to say that pursuing an NPI isn’t a laudable goal, but the challenges of creating safe and effective patient identifiers can’t be ignored, and it is critical that any effort made in this area be done carefully and with consultation of all parties involved in the healthcare ecosystem.
The industry also shouldn’t assume that a single identifier is ever going to be the end-all, be-all solution. Rather, it would be wise to treat an NPI as just another strong indicator of identity that, in conjunction with other demographics as well as biometric data, can assist with patient matching. The NPI could become an important piece of the overall identity puzzle, but it would not be definitive.
Instead of looking at a NPI as the definitive answer for solving our patient ID issues, the industry should focus on a pragmatic, multi-faceted approach to improve patient matching—one that moves beyond a reliance on probabilistic algorithms.
For example, it could supplement existing technologies that rely on demographic data with advances being made in biometrics as well as the ubiquitous use of smart phones. For instance, image processing and machine learning algorithms have made impressive advances in the last 10 years such that even the cameras of inexpensive smartphones can be used to capture accurate biometric signatures without the complexity of expensive and specialized equipment, like palm vein scanners. These innovations are mature enough to allow the broad adoption of facial recognition to the patient-matching equation.
As the industry embraces these and other technologies, it will increase its ability to accurately identify a patient at any time and in any care setting.